Longitudinal studies of children of alcoholics (COAs) have informed etiological theories about the development and course of substance disorders and documented COAs' pervasive risk for a variety of negative outcomes including substance disorders (Sher, 1991; West & Prinz, 1987). One potentially key mechanism explaining COAs' pervasive risk profile and individual variation in the course of substance outcomes is stress. We propose a developmental model of stress and substance outcomes that emphasizes the importance of time and timing in how stress impacts the life-course of COAs as well as the onset, escalation and deceleration of substance behaviors. Our method is a secondary analysis of COAs' early lifecourse which combines data from three, existing, methodologically rigorous studies initiated in childhood (Zucker & Fitzgerald, 1991), adolescence (Chassin, Rogosch & Barrera, 1991) and young adulthood (Sher, 1991). Despite the significant contributions of each study, differences among them in methodology, measurement and key questions of interest have not always facilitated comparison of results and substantive inferences about the life-course trajectories of COAs from early childhood into adulthood. Our proposal overcomes this issue through the pursuit of five aims: (1) to evaluate whether stress and the timing of stress events mediate the relation between parent alcoholism and individual variation in risk behaviors for substance outcomes over time, (2) to test whether five distinct mechanisms posited to underlie relation among stress, risk behaviors and substance outcomes are differentially salient across development among COAs and non-COAs, (3) to evaluate important moderators that either mitigate or exacerbate risk for substance outcomes related to the occurrence and timing of stress and risk behaviors in COAs and non-COAs, (4) to evaluate whether heterogeneity of risk exists within COAs due to comorbid disorders in alcoholic parents, gender of children and impaired parents, and patterns of drinking in alcoholic parents, and (5) to extend and disseminate existing longitudinal analyses that incorporate participants from multiple studies within a single analytic framework.